modify render delay

human/audio_inference_handler.py

@@ -83,7 +83,8 @@ class AudioInferenceHandler(AudioHandler):
                     logger.info('infer=======')
                     t = time.perf_counter()
                     img_batch = []
-                    for i in range(batch_size):
+                    # for i in range(batch_size):
+                    for i in range(len(mel_batch)):
                         idx = mirror_index(length, index + i)
                         face = face_list_cycle[idx]
                         img_batch.append(face)
@@ -130,6 +131,6 @@ class AudioInferenceHandler(AudioHandler):
         self._run_thread.join()
 
     def pause_talk(self):
-        pass
+        print('AudioInferenceHandler pause_talk', self._audio_queue.size(), self._mal_queue.qsize())
-        # self._audio_queue.clear()
+        self._audio_queue.clear()
-        # self._mal_queue.queue.clear()
+        self._mal_queue.queue.clear()

human/audio_mal_handler.py

@@ -76,6 +76,7 @@ class AudioMalHandler(AudioHandler):
                 print("AudioMalHandler start_idx222", start_idx + mel_step_size - start_idx)
                 mel_chunks.append(mel[:, start_idx: start_idx + mel_step_size])
             i += 1
+        print("AudioMalHandler start_idx333", i)
         self.on_next_handle(mel_chunks, 1)
 
         # discard the old part to save memory
@@ -105,4 +106,5 @@ class AudioMalHandler(AudioHandler):
         logging.info('chunk2mal stop')
 
     def pause_talk(self):
+        print('AudioMalHandler pause_talk', self._queue.size())
         self._queue.clear()

nlp/nlp_base.py

@@ -37,7 +37,7 @@ class NLPBase(AsrObserver):
 
     def completed(self, message: str):
         logger.info(f'complete:{message}')
-        self._context.pause_talk()
+        # self._context.pause_talk()
         self.ask(message)
 
     def ask(self, question):

nlp/nlp_doubao.py

@@ -33,7 +33,6 @@ class DouBao(NLPBase):
         logger.info(f'_request:{question}')
         logger.info(f'-------dou_bao ask:{question}')
         try:
-            '''
             stream = await self.__client.chat.completions.create(
                 model="ep-20241008152048-fsgzf",
                 messages=[
@@ -51,14 +50,14 @@ class DouBao(NLPBase):
                     self._on_callback(message)
             self._on_callback(sec)
             await stream.close()
-            '''
+
-            sec = "你是测试客服，是由字节跳动开发的 AI 人工智能助手"
+            # sec = "你是测试客服，是由字节跳动开发的 AI 人工智能助手"
-            sec, message = self._split_handle.handle(sec)
+            # sec, message = self._split_handle.handle(sec)
-            sec, message = self._split_handle.handle(sec)
+            # sec, message = self._split_handle.handle(sec)
-            if len(message) > 0:
+            # if len(message) > 0:
-                self._on_callback(message)
+            #     self._on_callback(message)
-            if len(sec) > 0:
+            # if len(sec) > 0:
-                self._on_callback(sec)
+            #     self._on_callback(sec)
         except Exception as e:
             print(e)
         logger.info(f'_request:{question}, time:{time.time() - t:.4f}s')

render/video_render.py

@@ -13,7 +13,7 @@ from human.message_type import MessageType
 
 class VideoRender(BaseRender):
     def __init__(self, play_clock, context, human_render):
-        super().__init__(play_clock, context, 'Video')
+        super().__init__(play_clock, context, 'Video', 0.035)
         self._human_render = human_render
         self._diff_avg_count = 0
 

tts/tts_edge_http.py

@@ -17,7 +17,8 @@ class TTSEdgeHttp(TTSBase):
     def __init__(self, handle, voice='zh-CN-XiaoyiNeural'):
         super().__init__(handle)
         self._voice = voice
-        self._url = 'http://localhost:8082/v1/audio/speech'
+        # self._url = 'http://localhost:8082/v1/audio/speech'
+        self._url = 'https://tts.mzzsfy.eu.org/v1/audio/speech'
         logger.info(f"TTSEdge init, {voice}")
 
     async def _on_request(self, txt: str):

utils/sync_queue.py

@@ -3,7 +3,7 @@
 import threading
 from queue import Queue
 
-'''
+
 class SyncQueue:
     def __init__(self, maxsize):
         self._queue = Queue(maxsize)
@@ -37,9 +37,9 @@ class SyncQueue:
 
     def size(self):
         return self._queue.qsize()
+
+
 '''
-
-
 class SyncQueue:
     def __init__(self, maxsize):
         self._queue = Queue()
@@ -55,3 +55,4 @@ class SyncQueue:
 
     def size(self):
         return self._queue.qsize()
+'''